Integrated parameter inversion analysis method of a CFRD based on multi-output support vector machines and the clonal selection algorithm

作     者：Zheng, Dongjian Cheng, Lin Bao, Tengfei Lv, Beibei 

作者机构：Hohai Univ State Key Lab Hydrol Water Resources & Hydraul En Nanjing 210098 Jiangsu Peoples R China Hohai Univ Natl Engn Res Ctr Water Resources Efficient Utili Nanjing 210098 Jiangsu Peoples R China Hohai Univ Coll Water Conservancy & Hydropower Nanjing 210098 Jiangsu Peoples R China 

出 版 物：《COMPUTERS AND GEOTECHNICS》 (计算机与土工学)

年 卷 期：2013年第47卷

页      面：68-77页

核心收录：

学科分类：0709[理学-地质学] 081803[工学-地质工程] 07[理学] 08[工学] 0708[理学-地球物理学] 0818[工学-地质资源与地质工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：National Natural Science Foundation of China [51079086, 51139001] Program for New Century Excellent Talents in University [NCET-11-0628] Special Fund of State Key Laboratory of China Fundamental Research Funds for the Central Universities Ministry of Water Resources Public Welfare Industry Research Special Fund Project 

主　　题：Concrete-faced rockfill dam Multi-output support vector machines Clonal selection algorithm Integrated inversion analysis 

摘      要：A geotechnical parameter identification method based on multi-output support vector machines (M-SVMs) and the clonal selection algorithm (CSA) is proposed. Using this method, the rockfill material parameters of a concrete-faced rockfill dam (CFRD) are identified. Based on the Taguchi design, some possible combinations of material parameters are generated within the admissible ranges of material parameters. Then, using these combinations, the displacement of all the observation points of the CFRD is calculated using the finite element method (FEM). Next, different combinations of material parameters are used as input, and the calculated displacement is used as output to train some M-SVM models to simulate the complex relationship between the material parameters and the dam displacement. Integrated inversion analysis, which takes the static and creep properties of rockfill material into account simultaneously, is implemented to achieve a comprehensive understanding of the mechanical properties of the rockfill material. The optimization problem corresponding to the integrated inversion analysis is solved by the CSA, which has global convergence and is robust. During the process of searching for optimal material parameters, the dam displacement is calculated by the M-SVM mapping instead of the FEM, which may greatly reduce the computation time. Based on the observed settlement and finite element model of the CFRD, the inversion analysis method described above is implemented. The results indicate that the parameter identification method for rock-fill material proposed in this article is accurate, has a fast convergence rate and can be applied in practical engineering applications. (C) 2012 Elsevier Ltd. All rights reserved.